Suggestions for Busy Winstat Servers

All of SSCC's servers (other than Silo) are very busy with everyone working from home, but Winstat is where people notice it the most just because that's where people work interactively. Some suggestions:

Please move all computationally intensive jobs to Linstat. The Linstat servers have either 36 or 48 cores and 768GB of RAM, much more than Winstat. On Linstat your job can keep running even after you log off, so there's no need to keep a session active. And since people rarely do interactive work on Linstat, it has less of an impact on others when the Linstat servers get busy. If you haven't used Linstat before, Using Linstat will get you started. With RStudio Server, you can run R jobs on Linstat without learning any Linux commands at all.

Please log out of Winstat when you're done using it, especially if you have Long Jobs access. That way the resources used by your session will be made available to others.

If you notice Winstat slowing down, signing out and signing back in will move you to the least busy server.

You are also welcome to use LabStats to access SSCC's lab computers remotely—that way you'll have a computer all to yourself.  The lab computers have a wide variety of statistical software installed and are a great alternative when Winstat is busy. Accessing SSCC Lab Computers Remotely has details.

Please Welcome David Thompson

We’re pleased to introduce our newest staff member David Thompson.

David Thompson has joined the SSCC as a Linux System Administrator. David joins the SSCC bringing 25 years of knowledge and experience in Linux system administration at UW Madison, most recently at the Waisman Brain Imaging Core and the Center for Healthy Minds. David also brings experience with a variety of scripting tools for Linux and macOS systems. David will work with Dan Bongert to administer SSCC's Linux servers, including the Linux computing cluster in Silo which was recently expanded thanks to our partnership with SMPH. When he’s not making technology go, he enjoys outdoor pursuits (biking/backpacking/canoeing/kayaking), music (both playing and listening), and brewing beers and meads at home. We are thrilled to welcome David to the SSCC team and look forward to seeing how his expertise will help shape the SSCC in the years to come.  

Fall Plans for the Computer Classroom and Labs

With the end of the pandemic on the horizon, we are taking reservations for the Computer Classroom (3218 Sewell Social Sciences Building) and the Mobile Lab for the fall semester. We are also pleased to announce that, with the approval of our Instructional Lab Modernization (ILM) grant, we will have new Mobile Lab laptops available for the fall semester. We are optimistic that the Computer Classroom and Drop-In Lab (4218 Sewell Social Sciences Building) will be able to operate as usual and at full capacity this fall, but safety remains our top priority and we will follow all guidance from the University.

We have not yet decided whether to open the Computer Classroom or Drop-In Lab during the summer, but we'd be very interested in talking with anyone who would use them.

If you are interested in using the lab during the summer or would like to make a reservation for the fall, please reach out to Caitlin Tefft ( We appreciate your continued flexibility as we make this next transition! 

GPU Computing at the SSCC

In January, SSCC put our first servers with GPUs into production: Linstat5 and Linstat6 in our regular environment, and LinSiloGPU001 and LinSiloGPU002 in Silo. GPUs, or Graphical Processing Units, were developed for rendering 3D graphics and contain huge numbers of small special-purpose processors. For example, the NVidia T4 GPUs in the new servers contain 384 "tensor cores" and 3,027 "shading units," among other things. But researchers soon realized that some research computing tasks could take advantage of all those processors and run much more quickly than by using smaller numbers of general-purpose CPUs.

Right now most of our GPU users run Python machine learning tools like PyTorch, but there are R packages that can take advantage of GPU—see the GPUs section of the R High Performance task view. Many Matlab functions will use GPU if you pass in a gpuArray, and others use it automatically—see the relevant Matlab documentation. SSCC's statistical consultants do not have experience with GPU computing (yet) but we encourage you to explore.

Data Carpentry session for the Social Sciences

May 12-19 the Data Science Hub is holding a Data Carpentry session intended for the social sciences. The curriculum is designed for researchers who will collect data in the field, and will introduce some computing tools that can be used to enter, clean, and analyze such data, namely, Excel, OpenRefine, SQL, and R. (SSCC's statistical consultants suggest that unless you're collecting your own data you can probably go straight to R, Stata, or something similar.) For details, visit the session web page. Registration opens April 5th.